Support Content-Encoding: x-gzip

[SpecLad]

Context

I'm using an API that returns a gzipped response, but the reported Content-Encoding is x-gzip rather than gzip, so urllib3 fails to decompress it.

According to RFC 9110:

A recipient SHOULD consider "x-gzip" to be equivalent to "gzip".

Therefore, I would expect e.g. Response.read(decode_content=True) to automatically decompress such responses.

Unfortunately, I can't find a public server that would produce such responses, so I can't provide a simple reproducer.

Alternatives

I can set decode_content=False and gunzip the content manually.

Duplicate

Has the feature been requested before?

Not that I could find.

Contribution

Would you be willing to submit a PR?

Probably not.

[sigmavirus24]

urllib3/src/urllib3/response.py

Lines 207 to 220 in 7d0648b

	def _get_decoder(mode: str) -> ContentDecoder:
	if "," in mode:
	return MultiDecoder(mode)
	if mode == "gzip":
	return GzipDecoder()
	if brotli is not None and mode == "br":
	return BrotliDecoder()
	if zstd is not None and mode == "zstd":
	return ZstdDecoder()
	return DeflateDecoder()

could be trivially updated to say, after we get past the MultiDecoder return do a if mode.startswith("x-"):\n\t\tmode = mode[2:] and then we don't need to worry about x-gzip or x-compress. And if any other encodings are eXperimental first, we can handle those backwards compatibly as well.

[SpecLad]

FWIW, I don't think it's a good idea to blindly assume that x-foo is the same as foo, since there is no guarantee that will be the case. Plus, the only registered x- encodings are x-gzip and x-compress (and urllib3 doesn't even support compress).

[sethmlarson]

@jeremycline has submitted a PR for this issue.